Amazon is retreating from the race to build the smartest AI. This week it closed its AGI Lab, a San Francisco team working on artificial general intelligence, The Information reported. The move was part of layoffs across Amazon’s AGI unit, Reuters first reported.
The lab was barely 18 months old.
Amazon set it up in December 2024 with several dozen staff. Its job was to make AI agents more useful to businesses. Now it is gone, and the people who built it are being let go. The full scale of the cuts is not yet clear.
Amazon frames this as focus, not surrender. “We’ve been building large AI models for several years, and it remains one of the most important things we’re working on,” a spokesperson said. The company is “sharpening our focus on the initiatives that matter most for customers”. That focus, it added, meant “eliminating some roles within parts of our AGI organization”.
The people at the top left first
The exits came before the layoffs. Rohit Prasad, the executive who oversaw Amazon’s AGI work, left at the end of last year. David Luan, who ran the AGI Lab, left in February. Amazon then folded its AGI work under Peter DeSantis, alongside its chip and quantum teams.
The pattern is not new. This is the latest in a rolling series of cuts since Amazon shed 16,000 jobs in January. Across tech, firms are trimming teams and pointing at AI. Losing two leaders and a whole lab in quick succession says a lot.
A frontier race Amazon never led
Amazon has spent a fortune on AI, but rarely at the frontier. Its own models have made little dent, and its real strategy runs through AWS and its huge AI spending on infrastructure. Where rivals build the smartest model, Amazon sells the picks and shovels, plus agents that plug into businesses. Its big model bet is external, with tens of billions poured into Anthropic.
There is a wider doubt in the background, too. A growing camp argues that today’s large language models are not the road to general intelligence at all. Whether Amazon shares that view or is simply chasing profit, the message is the same. The company that wants to power everyone else’s AI has stopped trying to build the cleverest brain itself. Its AI ambitions now point at selling AI, not inventing it.
